I made some more progress exploring the AD-25-500 detector. Now the detector is under vacuum, and to avoid noise problems, I put the entire CSA board in the chamber too. Only power, ground, detector bias, and pulse shaper output go outside the chamber.
CSA board (covered with black tape on both sides, to avoid shorts) on the bottom of the chamber, alpha detector mounted to top of KF25 flange over it):
Entire system (roughing pump, chamber, MKS 901p to confirm vacuum):
I ran the CSA output into the mic input of my laptop (after attenuating with a 10K pot), and ran the Theremino MCA software, using a fresh Am241 sample. The Am241 (ex-smoke detector) is about 1 inch from the detector. With normal air pressure, the detector gave around 160 counts/minute, with average pulse height around 350 mV. Under 50-100 micron vacuum, I estimate around 55k counts/minute, 1.4V average pulse height. The Theremino software count was much less - it's missing a lot of the pulses, either due to sampling rate issues, the software itself not keeping up, or user error (likely).
The laptop mic input samples at 96 kHz, so it's not really adequate for these pulses. One A/D sample is about the same width as the pulse itself (~15 us). You can see the resulting shape in the Theremino window - it's basically the impulse response of the sound ADC system rather than the original pulse as seen in the scope trace in the last post. But the overall sampled height seems to be roughly proportional, so it's not 100% worthless.
Anyway, still playing around with it, and open to further suggestions! I may try to hotwire the detector into the 7401 NIM alpha spectrometer module I got from Richard at the last HEAS. I don't have a NIM crate to power it up, so I'd have to hotwire that too.